Benjamin Rodenberg

> These could either live in `precice/precice` in `extra/performance/` or in a separate repository I think they should definitely live in this repository to make sure we detect performance regressions...

Provocative answer: If there is a valid use-case, there should be a test :smirk: I think I don't 100% understand the setup you are describing, but I'm also not very...

In the context of #1746 I see two options: 1. Live with the bug for now and comment out tests 2. Introduce Extrapolation I'm clearly in favor of option 1....

> In the current state, I would expect sampling past the last point to this to trigger an assertion. This would be the minimum. Yes. But might be possible to...

I added a `PRECICE_CHECK` to prevent increasing time window sizes from one iteration to the next via #1789. For the future, I created a draft PR containing the test that...

In https://github.com/precice/tutorials/pull/500 I implemented the higher-order time stepping scheme in a different way. Here, I basically provide a function handle to `readData` to the black-box time stepper. With this method...

Shifting it to `3.x.x`, because it is not critical in my opinion and the bug has always been there. Feel free to shift it back to `3.0.0`, if you are...

I think in both cases we are talking about data on the sending side that is not exchanged. So we need to refine this a bit. The problem in #1803...

Ideally we would be able to raise the error already during configuration time (not sure if this is possible). But a proper error at runtime also would be much better...

I currently don't have time to work on this issue, but I think it should be rather simple to fix this issue: * solution is already sketched above by @fsimonis...